Postfracture survival in a population-based study of adults aged ≥66 yr: a call to action at hospital discharge

Abstract

Postfracture survival rates provide prognostic information but are rarely reported along with other mortality outcomes in adults aged ≥50 yr. The timing of survival change following a fracture also needs to be further elucidated. This population-based, matched-cohort, retrospective database study examined 98 474 patients (73% women) aged ≥66 yr with an index fracture occurring at an osteoporotic site (hip, clinical vertebral, proximal non-hip non-vertebral [pNHNV], and distal non-hip non-vertebral [dNHNV]) from 2011 to 2015, who were matched (1:1) to nonfracture individuals based on sex, age, and comorbidities. All-cause 1- and 5-yr overall survival and relative survival ratios (RSRs) were assessed, and time trends in survival changes were characterized starting immediately after a fracture. In both sexes, overall survival was markedly decreased over 6 yr of follow-up after hip, vertebral, and pNHNV fractures, and as expected, worse survival rates were observed in older patients and males. The lowest 5-yr RSRs were observed after hip fractures in males (66-85 yr, 51.9%-63.9%; ≥86 yr, 34.5%), followed by vertebral fractures in males (66-85 yr, 53.2%-69.4%; ≥86 yr, 35.5%), and hip fractures in females (66-85 yr, 69.8%-79.0%; ≥86 yr, 52.8%). Although RSRs did not decrease as markedly after dNHNV fractures in younger patients, relatively low 5-yr RSRs were observed in females (75.9%) and males (69.5%) aged ≥86 yr. The greatest reduction in survival occurred within the initial month after hip, vertebral, and pNHNV fractures, indicating a high relative impact of short-term factors, with survival-reduction effects persisting over time. Therefore, the most critical period for implementing interventions aimed at improving post-fracture prognosis appears to be immediately after a fracture; however, considering the immediate need for introducing such interventions, primary fracture prevention is also crucial to prevent the occurrence of the initial fracture in high-risk patients.

Keywords: aging; fracture prevention; general population studies; osteoporosis; screening.

Graphical Abstract

Figure 1

Flow of participants in the fracture and nonfracture cohorts. The fracture and nonfracture cohorts used in this study were previously described.^,*Individuals assessed for inclusion in the fracture cohort sustained an index fracture between January 1, 2011, and March 31, 2015, at an osteoporotic fracture site (hip, clinical vertebral, pNHNV [pelvis, femur, sternum/rib/clavicle, and humerus/shoulder], and dNHNV [tibia/fibula/knee, radius/ulna, and wrist]) identified from hospital admissions, emergency room visits, and ambulatory care using International Classification of Diseases, 10th revision, Canada codes (Supplementary Methods). Individuals assessed for inclusion in the nonfracture cohort did not experience a fracture during and 5 yr before the index period and were matched to the fracture cohort on the following prespecified variables: sex, age group, rural/urban residence, and comorbidities associated with fracture risk. ^†Death date prior to index fracture date (n = 78) or non-Ontario resident (n = 123). dNHNV, distal non-hip non-vertebral; pNHNV, proximal non-hip non-vertebral.

Figure 2

Number of females and males with index fracture by fracture site category and age group. dNHNV, distal non-hip non-vertebral; pNHNV, proximal non-hip non-vertebral.

Figure 3

Survival probability in matched fracture and nonfracture cohorts. Dotted curves indicate nonfracture groups and are color-matched to their respective fracture cohort group. Number of patients at risk in nonfracture groups are labeled as “non-fx” and are listed directly below and color-matched to their respective fracture group. X-axis 0 indicates index date. Shaded regions represent 95% CIs. Dotted gray vertical lines indicate 1 and 5 yr of follow-up after index date. dNHNV, distal non-hip non-vertebral; fx, fracture; non-fx, nonfracture; pNHNV, proximal non-hip non-vertebral; Vfx, vertebral fracture.

Figure 4

Epanechnikov kernel–smoothed hazard functions in matched fracture and nonfracture females and males without CIs* and y-axis of 1.50 (A) and with 95% CIs (shaded regions) and y-axis of 5 (B). Lighter color curves indicate females, darker color curves indicate males, dotted curves indicate nonfracture cohort. X-axis 0 indicates index date. Shaded regions represent 95% CIs. Dotted gray vertical lines indicate 1 and 5 yr of follow-up after index date. ^*Based on Epanechnikov kernel smoothing of postfracture data over 6 yr of follow-up, large CIs were observed at the end of follow-up, and thus hazard function plots are provided without CIs in the A panel to support visual inspection of plots. dNHNV, distal non-hip non-vertebral; pNHNV, proximal non-hip non-vertebral.

Figure 5

Overall survival at 1 and 5 yr in matched fracture and nonfracture females and males. Nonfracture cohort data and labels are in italic font positioned directly below their respective fracture group and in dotted bars color-matched to their respective fracture cohort group. dNHNV, distal non-hip non-vertebral; fx, fracture cohort; non-fx, nonfracture cohort; OS, overall survival; pNHNV, proximal non-hip non-vertebral.

Figure 6

RSRs at 1 and 5 yr post-fracture in females and males. dNHNV, distal non-hip non-vertebral; RSR, relative survival ratio; pNHNV, proximal non-hip non-vertebral.